Pulmonary nodule detection: low-dose versus conventional CT.

Abstract

To quantitate the effectiveness of low-dose computed tomography (CT) in the identification of pulmonary nodules while controlling for anatomic nodule characteristics and to establish what factors lead to reduced diagnostic sensitivity at low-dose CT. Each of six participating radiologist independently rated 200 image panels by using a four-point confidence scale. Conventional images were obtained at 200 mAs; low-dose images were obtained at 20 mAs. To fully control their characteristics, nodules were simulated with a given diameter, shape, and section thickness while preserving the resolution, noise level, and reconstruction artifacts of the original images. Panels were matched so that nodules on low-dose and conventional images had equivalent sizes, locations, and relationships to blood vessels. Among 864 positive panels, 259 (60%) of 432 low-dose panels and 272 (63%) of 432 conventional panels were correctly interpreted (P = .259). Lowering the x-ray dose significantly reduced the detectability of peripheral nodules (P = .019) and nodules separated from blood vessels (P = .044). Surprisingly, 3-mm nodules were detected with approximately equal sensitivity (P = .181) at conventional and low-dose CT. The specificity of low-dose images was 88% (148 of 168 panels) versus 91% (153 of 168 panels) for conventional images (P = .372). Low-dose CT is acceptable for pulmonary nodule identification, making it suitable for primary screening. These results confirm the strong effect of size, location, and angiocentricity on the sensitivity of nodule detection with conventional CT.